mirror of
https://github.com/stablyai/orca.git
synced 2026-11-04 11:07:42 +00:00
5ffc8d7972c7bef9788cffd6f90b1e00d1e014ec
13274
Commits
| Author | SHA1 | Message | Date | |
|---|---|---|---|---|
 Brennan Benson
|
5ffc8d7972 |
Keep Chat UI users with a Terminal default on the terminal, with a chat mode switch in the upgrade tip (#26915)
* Keep Chat UI users who opened new tabs in the terminal on the terminal Before the chat upgrade, Experimental Chat UI had a separate Default view choice, and it defaulted to Terminal. The upgrade retired that choice so Chat UI on now means new agent tabs open as chat, which moved people who had kept the terminal default into chat with no obvious way back. On load, a profile saved by a pre-upgrade build with Chat UI on and Default view not set to Chat now gets Chat UI turned off, and is left out of the one-time native chat upgrade tip. The old key is stripped on the first save, so the migration runs once and a later opt-in sticks. * Update the saved-settings test for Chat UI on with Default view Terminal * Show the upgrade tip to terminal-default users, with a switch to turn on chat mode People who had Chat UI on with new tabs opening in the terminal now keep the terminal on upgrade, but they still need the tip's directions for moving between chat and CLI. They now get the same upgrade tip, plus a note that new agent tabs still open in the terminal and a "Turn on chat mode" switch. The switch shows when Chat UI is off as the tip opens and stays put while it is flipped. The tip is taller in that case so nothing scrolls. * Decide the chat mode switch from the saved tip audience, not the live setting The upgrade tip's 'Turn on chat mode' switch showed whenever Chat UI was off when the tip opened, so a chat-view member who had turned Chat UI off also got it. Main now answers with one tip variant (none, standard, keep-terminal) from the saved audience, and only keep-terminal shows the note and switch. Also moves the feature tip panel styling into a DialogContent surface variant and drops the Label restyle, and tests that the switch stays put after flipping. * Let the upgrade tip decide the chat mode switch from its variant The modal now passes the saved tip variant to the dialog, which decides the switch itself, so the dialog tests cover that decision. |
||
 Jinwoo Hong
|
d358e5973c |
feat(relay): one-cell flag write workflow with dry run, compare-and-swap and read-back (#26837)
* feat(relay): one-cell flag write workflow with dry run, compare-and-swap and read-back Operate Relay Production Cell Flags writes cells/<cellId>.json for one cell: it checks the origin answers as that cell with a flag channel, refuses a stale expected generation, writes with ifGenerationMatch, stamps the actor and run on the object, and waits up to 15 s for the cell's runtime status to report the new generation applied. The deploy Workload Identity condition admits the new workflow. * fix(relay): flag workflow retries a transient 5xx, shows the plan before the generation check, and names the failure Runtime-status reads go through the shared one-retry admin helper, and a failed read-back poll no longer ends the wait early. The plan line is logged before the generation check, a failure is logged as one JSON line the job summary carries, and the object and summary name the triggering actor. |
||
|
Jinwoo Hong
|
1472b6803f |
feat(relay): Cloud Storage bucket for the per-cell runtime switch files (targeted apply only) (#26835)
* feat(relay): Cloud Storage bucket for the per-cell runtime switch files <project_id>-relay-control, versioned, uniform access, public access prevented. Cells and directors read; only the flag workflow's deploy identity writes. Apply with a targeted apply, never a root plan. * fix(relay): scope the switch-file writer to cells/ with a minimal custom role; soft-delete keeps history The deploy identity is shared, so its grant now reaches only cells/<cellId>.json and carries only get, create and delete. A 90-day soft-delete window keeps every deleted generation recoverable, since overwrite needs delete and delete also reaches old versions. * refactor(relay): drop the speculative directors.json read grant from the control bucket |
||
|
Jinwoo Hong
|
37d2154765 |
feat(relay): read-only seat feed on cells for the directors' shadow directory (#26803)
* feat(relay): read-only seat feed on cells for the directors' shadow directory Cells keep an in-memory log of host joins, leaves and drain notices and serve it at GET /v1/admin/cell-seats, authenticated with the directors' existing rehome identity. Nothing reads it yet. * test(relay): read the seat feed through its published shape so the tests typecheck * fix(relay): serve the seat feed's full snapshot from the change log's own seats counts.controls drops when a socket starts closing, but the leave lands only at its close event, so a full snapshot built from open sockets could disagree with a reader that applied every change. The log now keeps its own seats under the readers' generation rule and reports counts.seats. * fix(relay): report auth-expiry drain-only and its refresh as seat-feed changes The heartbeat's auth-expiry drain-only and an auth refresh back to active changed session state without a feed entry, so a reader's seat state drifted from the cell's. Both now go through the log; the refresh is a new 'active' kind. * refactor(relay): cut unread seat-feed fields; verify a feed poll once fixLevel, counts.splices, counts.enforcedUnits and counts.hardCap had no reader, so they leave before they become a cross-version contract. The admin middleware no longer verifies /v1/admin/cell-seats polls, which the route already gates on the rehome identity. |
||
|
Jinwoo Hong
|
dc69efa113 |
fix(relay): a database fault reaching unhandledRejection no longer kills the process (#26817)
* fix(relay): a database fault reaching unhandledRejection no longer kills the process
A rejection is fenced only when the relay's database layer threw or
rethrew it and it is a pool acquire failure, pg's read timeout, a dropped
connection, or a server error in classes 08/23/40/53/55/57/58. Everything
else stays fatal and is named by an orca_relay_process_fatal line first.
Every pooled client also keeps a permanent error listener. Fix level 2.
* fix(relay): name class 23 in the fence comment; move the fix-level bump to the read-timeout fix
R1 alone does not close the lost-reply stall, so the level rises with R2.
The fix-level comment now describes the fixed alert floor.
* fix(relay): survive and answer 503 for 25P03, the idle-in-transaction timeout a VM stall triggers
A whole-VM database stall of 6-7 s outlasts the 5 s idle-in-transaction
limit for a backend caught idle mid-transaction, and the server ends that
session with FATAL 25P03 on resume. The fence now survives that one class-25
code (the rest of class 25 stays fatal), and routes answer it as transient.
* style(relay): one transient SQLSTATE per line
* fix(relay): fence a socket reset or broken pipe the database layer rethrew
pg 8.22 rejects the active query with the raw socket error ('read
ECONNRESET'), which the message-only check let through to a fatal exit:
the 09-28 crash shape. Match ECONNRESET and EPIPE by code, gated by the
database-layer mark.
|
||
|
Jinwoo Hong
|
80d7e2dded |
fix(relay): spread drain placement over a load band instead of one empty cell (#26802)
* fix(relay): spread placement over a load band so a drain does not land on one empty cell Placement took the single least-loaded cell, so a just-rolled cell took a whole drain and every follow-up serialized on its relay_cells row. Directors now pick at random among cells within 5% of capacity of the minimum, and skip a cell that already took over 40% of its pool's placements in the last 10 s (floor 4). It only reorders: the least-loaded cell is the fallback, the hard cap and region preference are unchanged. * fix(relay): size the load band's intake window for the 1,200 s drain pace At ~0.5 picks/s per director, a 10 s window with a floor of 4 let the floor decide and the empty cell took ~65% of a c27-sized drain. A 60 s window and a floor of 2 keep it at 40% at both the 300 s and 1,200 s paces across five directors. Adds a once-a-minute orca_relay_placement_load_band line (picks, budget skips, fallbacks, band size, picks per cell). * chore(relay): log the load-band summary at info, not warning |
||
  
|
62878c3b0c |
fix(folders): folder workspaces can answer "paired server" (#26958)
* fix(folders): folder workspaces can answer "paired server" * fix(folders): name a server folder's nested target only when its id is unambiguous * fix(folders): read the folder group only when the answer needs it --------- Co-authored-by: m4air <m4air@Mac.localdomain> Co-authored-by: m4air <m4air@m4airs-MacBook-Air.local> |
||
|
Jinwoo Hong
|
4a964633f4 |
fix(automations): stop offering Blank Terminal where it cannot be chosen (#26961)
Blank Terminal is now opt-in on AgentCombobox. Only the new-workspace composer opts in; automations, source-control agent actions, and chat continuation offer agents only. The props are a discriminated union so agent-only callers' onValueChange never receives null. Fixes STA-9868 |
||
|
OrcaWinandm4air
|
b8a0554357 |
refactor(runtime): runtime git commands use a provider for every host (#26965)
requireRuntimeGitProvider now returns a provider for the local host too (the local git provider, built per call from the worktree's WSL options and shared link paths), so the runtime git commands drop their "SSH provider, else local free function" branches. Push-target materialization stays host-specific and moves into one helper. Branch-diff keeps its two arms because the local git function returns one pinned diff, not the contract's list. Co-authored-by: m4air <m4air@m4airs-MacBook-Air.local> |
||
|
Brennan Benson
|
2356baa181 |
Open a chat visual in its own tab instead of the sidebar (#26728)
* Open a chat visual in its own tab instead of the sidebar * Keep chat visual tabs alive across tab switches and finish editor-tab consolidation - Render chat visual tabs from a worktree-level retained overlay so switching tabs or moving the tab to another split never reloads the page. - Reopening a visual focuses it in whichever split holds it, takes its newest title, and opening from the floating panel no longer takes the main window's editor selection (check-details opener uses the same helper). - Visual tabs return to the reopen-closed stack and show in the tab palette. - One editor-family list and one virtual-tab predicate: no Copy Path, Copy Relative Path, Reveal or git status decoration for check-details and chat visual tabs. * Give chat visuals one open path and re-read them when shown - Reopen-closed (Cmd/Ctrl+Shift+T) of a chat visual goes through the visual opener, so it focuses the visual in whichever split holds it and keeps its id when a runtime environment is active. - A visual tab re-asks its host each time it is shown or reopened; an unchanged revision keeps the page as is, a changed one shows the new page, and an unavailable tab recovers. - The tab palette no longer shows or matches a virtual tab's internal id; the new-tab search shows the chart icon for visuals. - Cmd/Ctrl +/- uses app zoom and Alt+Z leaves word wrap alone while a virtual tab (CI details, chat visual) is active. - A tab dragged over a visible visual passes through the frame, using the existing drag passthrough. * Slim the chat visual tab back to the minimum Remove the retained overlay, the visibility re-read, the zoom and word-wrap rerouting, palette inclusion, reopen-closed routing and the check-details opener change. The visual tab renders through the editor area again, stays off the reopen-closed stack, and keeps: one tab per visual across splits, no main-window selection change from the floating panel, the newest title on reopen, and no Reveal or git status letter on virtual tabs. * Re-read an open chat visual when it is opened again; no watcher for virtual tabs - "Open in new tab" on a visual whose tab is already on screen in another split now re-asks the host, so a rewritten visual shows its new page. - CI details and chat visual tabs no longer start a workspace file watcher. - Drop the unreachable chat-visual case from the restore-time transient check, use the shared virtual-tab check in the header path, and say why visuals stay off the reopen-closed stack. * Let tab drags pass over chat visuals; hide path copies on virtual tabs - A chat visual's frame joins the existing tab-drag pointer pass-through, so a tab dragged over a visual (in its tab or inline in a chat) can drop or split there instead of the frame swallowing the drag. - The tab menu no longer offers Copy Path / Copy Relative Path for CI details and chat visual tabs, whose path is an internal id; no stray separator. * Reuse the hover icon for editor tab icons, so a chat visual shows its icon in both |
||
 Kelvin AmoabaandBrennan Benson
|
2cb2bef67e |
feat(native-chat): highlight code blocks and shell commands (#26343)
* feat(native-chat): highlight code blocks and shell commands Uses Shiki rather than the highlight.js already in the preview: it can resume mid-block while a reply streams. * fix(native-chat): leave the tool row command unhighlighted * refactor(native-chat): highlight code through a shared TextMate interface Chat code now goes through one renderer highlighting interface built on Shiki core and the Oniguruma WASM the editor already loads for Nim and Typst, with VS Code Light+/Dark+. Lines that are too long or too slow leave the rest of the block plain, failed grammar loads retry from a mounted block, every block shares one time slice, finished-block caching counts retained text, each theme keeps its own formatting, Vue template expressions are colored, and tool-row commands take the grammar of the shell they ran in. * chore(deps): drop pnpm self-install entries from the lockfile Keeps the lockfile change to Shiki alone; the earlier commit picked up the @pnpm/exe entries pnpm 12 writes when it runs in a fresh checkout. * test(native-chat): give cold grammar loads time in highlighting tests The first colored block in a test file loads its grammar and the regex engine, which took over a second on a busy CI runner. --------- Co-authored-by: Brennan Benson <79079362+brennanb2025@users.noreply.github.com> |
||
|
OrcaWinandm4air
|
470f685b80 |
refactor(providers): local git provider behind the host router (#26955)
* refactor(providers): local git provider behind the host router Adds createLocalGitProvider, which binds this machine's git free functions to the IGitProvider contract with the admission tiers and WSL options the local call sites use today. The router's local git route now carries it as a per-call factory. No caller changes yet. * refactor(providers): read shared link paths lazily in the local git provider --------- Co-authored-by: m4air <m4air@Mac.localdomain> |
||
|
Brennan Benson
|
dfffb302f8 |
Start orchestration workers as native chats for every agent that has native chat (#26776)
* fix(orchestration): start native-chat workers for every registered native-chat agent * chore(orchestration): correct the restored worker agent fallback comment |
||
|
Brennan Benson
|
8defd660c7 |
A new Pi chat names the model it will run on its first frame (#26821)
* refactor(native-chat): one rule for what a no-pick session says of its configured default ACP's rule (only a new session that restored no model pick and was sent none runs its config's default; an effort it picked says nothing) moves beside the catalog entry types so other agents' adapters report through the same step. An empty listing (a signed-out agent) now says nothing instead of "no listed model", so it can never retire a saved default. * feat(native-chat): check Pi's project config before naming its default model Pi reads `.pi/settings.json` in the chat's folder and loads project extensions from `.pi/extensions` (and the `extensions`/`packages` lists). A settings file that sets defaultProvider, defaultModel, defaultThinkingLevel, modelThinkingLevels or enabledModels, loads extension code, or can't be parsed counts as the project picking the model; extension code may register providers or set the model itself. Key names come from Pi 1.0.4's own settings manager and docs, read without running Pi. * feat(native-chat): a new Pi chat names the model and thinking level Pi starts on Pi's model listing marks no model as the one it runs, so a new Pi chat with no saved pick showed "Model" and then changed to the model its session reported. A new Pi session (not one resumed or forked from a saved conversation) that restored no model pick and was sent none now reports the model and, unless it picked one, the thinking level it started on, through the same host step as Grok, OpenCode and OMP. The host saves them as the account's default behind the project-config check, so the next new Pi chat names them on its first frame. A signed-out Pi, which lists no model, saves nothing. * fix(native-chat): a newer, coarser effort menu keeps the effort a chat ran The saved list takes each model's effort menu from the newer listing. Pi's `--list-models` menu stops at high, while a chat's own menu adds xhigh or max for models whose thinking map allows them. After a background re-list, a learned xhigh was no longer in the menu, so the first frame dropped it. When the configured effort is only in the other listing's menu, that menu is kept. * fix(native-chat): Pi teaches only the model and thinking level it started on A Pi chat re-derived its configured default from Pi's current state on every options read, so a model an extension switched to mid-chat (extensions can call setModel) was saved as the account default, and the next new chat named the wrong model. The choice is now computed once, after start-up restores the chat's saved picks, and every later read reports that snapshot. A pick made in the chat no longer matters to it. Tests cover an extension switch, a pick made in the chat, a forked chat, and a learned xhigh surviving a newer re-list. |
||
|
Brennan Benson
|
573ff0bca3 |
Save a new chat's first message during create (#26332)
* fix(native-chat): the host owns the send queue; the window keeps no saved outbox The desktop kept each structured chat's unsent messages in localStorage and sent them in order, so one message whose fate was unknown froze every later send, Retry dropped it silently, failed sends could not be discarded, and an offline chat could send hours later. The host already records every message and owns the queue; the window now only sends. - One in-memory sender for composer, launch prompts and messages sent from outside the chat. One send in flight per chat; a transport failure resends the same id for up to 30 s; a refusal that proves nothing was recorded puts the text back in the composer with the reason; a send that went out and was never answered shows an in-doubt line with Send again and holds nothing up. - The host's "unknown" rows get the same in-doubt line and Send again, from the journal, in every window. - A message an older build left in localStorage is never sent: the host's conversation outline decides what goes back to the composer, and the copy is deleted once that is saved. * fix(native-chat): send through the structured chat RPC wrapper, with its timeouts The sender called the runtime RPC directly, skipping the per-method timeouts every other structured chat call gets. Only a remote host's request skips the compatibility check the sender already ran. * fix(native-chat): an unconfirmed send goes back to the composer, with no new row line The common pattern draws nothing extra on a message whose delivery is in doubt once its turn is over, and puts a failed send's text back in the composer with the reason. So a send nobody answered in time, or one a host answers in a way that proves nothing, goes back to the composer worded as unconfirmed, and a host "unknown" row shows nothing extra. Removes the in-doubt phase, Send again, and its strings. A host's made-up row for an id its journal lost now reads as unconfirmed, not as recorded, so that message comes back instead of vanishing. * test(native-chat): pin that nothing resends a send given back as unconfirmed * fix(native-chat): sends survive a tab close, never resend after a Stop, and keep remote images - A normal tab close lets sends on their way settle; what the host never took goes back to the conversation's draft. Only a cancelled launch or a worktree purge drops them. - After any Stop, a send already on its way is never sent again under its id: a doubtful answer, or a resend that was due, hands its text back worded as unconfirmed. - A returned image keeps the SSH connection it lives on; the connection never goes to the host. - An older build's saved message the host recorded and then rejected is left to the host's own row, never handed back. * fix(native-chat): a Stop or a tab close never resends a send already out, and a kept card is the card's A Stop that landed while a same-id resend was being readied (its timer fired, its request not out yet) let that resend go out after the Stop. The sender now tracks whether a request is awaiting its answer: a Stop hands back every send between attempts as unconfirmed, lets one whose request is out settle from its answer, and never issues a request after it. A normal tab close withdraws the same way instead of doing nothing, so nothing more goes out and nothing is dropped. A send the host rejected but kept as a card (keptAsQueuedMessageId) is the card's, from its reply or the journal, and never goes back to the composer. Adds the freeze tests: a send whose fate is unknown holds later sends no longer than its deadline, and one the host can neither confirm nor deny releases the next at once. * fix(native-chat): read a resend's turned-away call or reused id as unproven Ports the send-answer proof contract. A call the host turned away before running it (method_not_found, invalid_argument, unauthorized) proves only that this request wrote nothing, so it reads as never sent on a first attempt only; on a resend an earlier attempt may have landed, and it goes again under the same id. A resent id the host says was already used for other content (messageIdReused) proves nothing either, like an expired or conflicting id. Pins the rest of the contract: any row the host returns is its own, a thrown error is no answer whatever its code, and an older build's refused, held or outlived-Stop copy is handed back once and never sent. * refactor(native-chat): type the structured composer's send with its attachment type Keeps NativeChatStructuredSession.tsx within the file length limit. * refactor(native-chat): drop the kept-card guards the sender never needed A kept send's row is a rejection that was never a Stop's, so the sender already reads it as recorded, from its reply or the journal. The test that pins it stays; the two extra checks only covered a kept row that is also a withdrawal, which the host never writes. * test(native-chat): route launch tests' sends through the client wrapper the sender calls The sender sends through callStructuredAgentSession, but these launch tests replaced that module with a factory that answered nothing (and still named a probe that no longer exists), so every launch prompt resent until its 30 s deadline and the tests timed out. Each factory now hands sends to the runtime RPC mock the tests already answer, and expectations of a local send no longer ask for the remote-only compatibility option. * fix(native-chat): hand a message back without importing the composer's attachment hook The worktree purge reaches the launch prompt, which hands text back, and the attachment hook's imports reach the store. A test that builds the real store behind a mocked one then waited on itself and hung. Handing back now writes images to the draft store directly, as the hook's helper did, and a test pins that each returned image keeps its SSH connection. * fix(native-chat): one send per chat, with no line of sends behind it A chat with a send out took further messages into an in-memory line and sent them one by one. A message typed behind one in doubt then hit its own 30 s deadline and came back as not sent without ever going out. Now, as the common pattern does, a chat takes one send at a time: while it is out, Send is disabled and Enter leaves the text in the box. The sender refuses a second send instead of lining it up, so the 30 s deadline always runs from the send itself. A Stop or a tab close stops the one send: settled from its answer if its request is out, handed back as unconfirmed between attempts, or silently if it never went out. Notes sent from outside the chat while its send is out stay with their sender (not ready); a launch prompt that meets the person's own first message waits in the composer instead of being lost. * fix(native-chat): give a Stop-withdrawn note back to its chat once its notes were cleared Notes sent from outside a chat clear once their message is recorded. A message the host recorded as pending and a Stop then withdrew came back only to its sender, which had already let go of it, so the text was lost. The chat's draft now takes it, as an earlier build's outbox did. * test(native-chat): type the composer-actions probe without a cast * chore(native-chat): drop outbox wording left in comments and an empty locale group * fix(native-chat): pace failed checks, keep the host's reason, and hold the chat for its launch prompt - A send whose checks failed before its request went out (an unreachable or incompatible host, an unreadable history) was tried again at once, about a thousand times a second for 30 s. Attempts are now paced by the attempts made, whether or not their request went out. - A host that refuses every resend by throwing (native chat turned off, a journal that won't open) gave back only "couldn't confirm". The host's reason now comes first, still without claiming not sent. - A launch's prompt now holds the chat's one send from the click, drawn as sending, so a message typed while the chat starts can't overtake it; it goes out once the chat exists, and a cancelled launch frees it. - Notes whose send nobody could confirm say so instead of "did not accept", and notes launched into a new chat let go of their text once that chat's composer holds it. - An open chat keeps drawing a recorded send until its row arrives, so a reply that beats the history no longer makes the message flicker out. * fix(native-chat): hold a chat's sends until it has started, and send a failed chat's message with its restart A message sent to a chat still starting went out at once to a host that had no record of the chat yet, read for a fence it could not get, and came back as not sent. A message sent to a chat whose start failed restarted it but no longer went with the restart. While a chat starts, Send stays off and Enter leaves the text in the box, as with a send already out. A text message sent to a chat whose start failed restarts it and goes as the restart's first message, through the same staged-prompt path a launch prompt takes: it holds the chat's one send slot, drawn as sending, until the chat exists, and comes back to the composer if the restart fails again. Notes wait while a chat starts and ride a failed chat's restart, keeping their text until it is sent. * chore(native-chat): test the queue request, rejection words and card hand-offs; drop an unused clear - Pin which sends ask the host to queue, the moved rejection wording, and that a queue send whose card was handed off and then refused or withdrawn, or whose replay names a withdrawn card, is never handed back. - The send-at-most-once gate now says what the desktop promises after a reload: it never sends the id again, and hands the text back. - The legacy read names when it goes, and drops the notice clear nothing called. * fix(native-chat): keep a sent launch prompt's entry, and give back at once what can't go out - Cleaning up a launch prompt released its send slot even after the prompt had gone out, which deleted the entry the sender keeps once the host records it. A launch prompt recorded as pending and then withdrawn by a Stop was lost from both the chat and the box, and an open chat dropped the new chat's first message before its row arrived. The slot now gives back only a reservation that was never sent. - A send stopped before its request went out by something trying again won't clear (this client and the server can't talk, or the host refused the history read) kept Send off for 30 s and then said Orca couldn't reach the agent. It now comes back at once with its own cause: not sent, since nothing went out, or unconfirmed if an earlier attempt did. Transport errors keep the paced retries. * fix(native-chat): notes keep their own text through a new agent's launch Notes sent to a New agent whose start failed came back to the notes and also sat in the new chat's composer, so sending both delivered the text twice. The notes keep their text until it goes out (they hold it from the click), so the launch now says so and no composer gets a copy, on a failed start or a refused prompt alike. The tests that asserted the copy in the composer pinned the old double ownership and now assert the notes are its only owner. Notes that rode a failed chat's restart and ended unconfirmed now say Orca couldn't confirm them, as notes sent directly do, instead of that the agent did not accept them. * chore(native-chat): the send-once gate says what the desktop keeps across a reload The desktop keeps a send's id in memory only: a send still unsettled at a reload or crash is not resent and not handed back. The coverage notes no longer credit the renderer tests with durable identity across a remount. * fix(native-chat): say once why notes sent to a new agent did not go Since notes keep their own text through a new agent's launch, a prompt the host refused, nobody could confirm, or that found the chat's send taken came back to the notes with nothing said anywhere: the chat shows no notice for text its caller keeps. The notes menu now reports it once, with the toast a send to an existing chat already uses: not accepted, couldn't confirm, or not ready. A start that failed still says so in the new chat instead. * fix(native-chat): retry a history refusal the host says clears, and name it at the deadline A send stopped before its request went out came back at once for any refusal of the history read, including ones the host names as clearing (its journal briefly unavailable, a chat detached while the host quits), so the automatic retry was lost. Only a version mismatch and refusals that won't clear come back at once now; the rest go again on the paced schedule, and if the deadline still finds nothing sent, the words are the host's refusal rather than Orca couldn't reach the agent. * feat(native-chat): a send makes one request, and nothing ever sends it again The desktop resent a message under its own id for up to 30 s when the answer was lost. A send now makes exactly one request, as the common pattern's clients do: - An answer that is lost, dropped or proves nothing hands the text back at once with "couldn't confirm… check the chat". - A failure before the request goes out (the environment check, the history read for the fence, a refused connection) means nothing went out: the text comes back at once with its own reason, or as not sent. - The 30 s cap stays on the one request, so a host that never answers can't hold Send. Nothing ever resends, so nothing can go out after a Stop: a Stop only takes back a send still in its pre-send checks. The resend state goes with it (tries, generation, awaiting, stopped, the resend timer, the send-answers-proof probe, and the first-attempt/resend split in the evidence), and the send-once gate says the desktop never resends. * fix(native-chat): notes keep the chat's line, and a send held behind a rewind says it was not sent - Only a send whose text belongs to the chat's composer clears the chat's line. Notes sent from outside the chat no longer wipe a "couldn't confirm... Check the chat" line that explains text already back in the box. - A send the host turns away behind a rewind it could not confirm went back as "not sent" but said "couldn't confirm what happened. Check the chat". It now gives the rewind's reason and says the message was not sent. - Reliability gate names the single-request test and records a fresh evidence run; the sender test drops its leftover resend mocks and a duplicate Stop test. * fix(native-chat): a send ends even when putting its text back fails If writing the returned text into the chat's draft threw, the send never settled: it stayed "sending", the chat refused every later send until a reload. The send now always ends after a hand-back, the failure is logged, and the chat's line adds "Couldn't save your message." * fix(native-chat): a message typed during /clear stays in the box and follows the chat A /clear moves the chat to a new conversation, and its host refuses any send while it runs. A message sent then went out, came back refused into the old conversation's draft with its line, and vanished when the chat moved on: the box and the line now belong to the new conversation. - A /clear holds the chat's one send slot while it runs: Enter does nothing, Send shows busy and the text stays in the box, as for a send that is out. Notes sent from outside get "busy". - Once it moves the chat, the old conversation keeps taking no send until the view leaves it, and what is left of its draft moves into the new conversation's draft, after anything there: when the composer's /clear settles, and again when the view moves. * fix(native-chat): no "Send message?" or queue clear while the chat's send is out While a send is out (or a /clear runs) the chat takes no message, yet Enter over a held queue still opened "Send message?", and Clear queue deleted every card before its message was refused. Enter now does nothing there and the text stays; Clear queue re-checks and deletes nothing if a send went out after the question opened. * refactor(native-chat): take the /clear draft carry out of this change Moving the old conversation's draft into the one a /clear replaces it with fixes a bug main has too (text left in the box during a /clear stays with the old conversation), so it goes in its own change. Kept here: a /clear holds the chat's send slot while it runs, and the conversation it moved away from takes no send until the view leaves it. * fix(native-chat): a /clear's hold on sends always ends - A view that unmounted while a /clear that moves the chat was out left the old conversation holding its sends until a reload: the late reply kept the hold for a view that was gone. The reply now releases it. - The local call for a conversation command has no deadline of its own, so a /clear that never answers kept Send off for good. The hold now also ends at the command's deadline (195 s, the one the remote call already uses), whichever comes first. * test(native-chat): opening a chat an older build left stuck; hand back its copy in send order Pins, through the real chat hook, sends, legacy recovery and draft store, what opening such a chat does: the queued messages behind a message the host recorded in doubt come back to the composer once, in order, with the "not sent" or "couldn't confirm" wording; the chat is free to send under its read's fence; nothing happens while the chat has no fence here. The legacy reader now hands entries back in the order they were sent (queuedAt), as the older build's reader did, instead of array order. * fix(native-chat): a send this window turned away for a re-paired server comes back as not sent A managed server's update rotates its pairing, and this window's main process then answers the next call itself, before forwarding anything, with runtime_environment_changed. The send read that thrown answer as proving nothing, so the person was told Orca couldn't confirm a message that never left. It is now handed back as not sent, with the reason. Every other thrown answer still reads as unconfirmed once the request may have gone out. * Save a new chat's first message during create * Align opening-message types with the user-message wire * Keep chat publication and cleanup outside store initialization * Complete the create refusal fixture ledger * test(native-chat): prove a created first message waits for the agent's start and is handed over once * refactor(native-chat): keep the host and session hook under the line limit after the main sync * test(native-chat): a chat created with its first message and no pick still teaches its agent's default After the main sync, a created chat's agent starts from the delivery loop rather than the create call. Grok (through the real ACP adapter and model catalog) and Claude (through the real adapter) each still report the model their own config resolved to once the chat reads its options. |
||
 Neil
|
35b145dbc8 | [CI] Run Activity tests together and replace the real idle wait (#26949) | ||
|
OrcaWinandm4air
|
6a929b2023 |
fix(hosts): host registry stops minting client SSH hosts from a server's nested target (#26952)
Co-authored-by: m4air <m4air@Mac.localdomain> |
||
|
Jinwoo Hong
|
8b2a5e45f1 |
Re-land Claude per-account folders (#24434, #26405) with the upgrade and proxy fixes (#26801)
* Revert "Revert Claude per-account folders (#24434, #26405) before release (#26638)"
This reverts commit
|
||
|
OrcaWinandm4air
|
ce9eb662bb |
refactor(renderer-ui-store): clean up Phase 3-3.8 code (#26903)
Share one bounded generation map across SSH, runtime SSH, slug resolution and session-tab tracking; reuse the SSH target generation helpers; replace the removed-flag chain in SSH target cleanup with a patch accumulator; derive purge recency targets from the rows the drop already removed; share the add-repo stale guard and runtime target; share the host-setup availability rule; drop the dead sync session persister; reuse host-id constructors in the file explorer; trim remote browser stream comments. Co-authored-by: m4air <m4air@Mac.localdomain> |
||
|
Jinwoo Hong
|
65b2fcf89a |
Start AI commit messages from the same shell environment chats use (#26808)
* Start AI commit messages from the login-shell environment chats use A Dock-launched Orca only takes PATH from the user's shell, so variables exported in the shell rc (proxy base URL, auth token) never reached the commit-message agent even though chats and terminals saw them. Local host generation now starts from the same cached login-shell base env as chats, with account overrides applied on top; WSL and SSH paths are unchanged. * Prefer the login shell's config directory over Orca's inherited one for commit messages |
||
|
OrcaWinandm4air
|
6dab0cb7fe |
refactor(ipc-persistence-shared): clean up Phase 3-3.8 code (#26902)
- Install store domain contexts with one generic symbol copy - Share one import-source classifier and one copy-name deconflicter across local, SSH and runtime-upload imports - Collapse kill.ts synthetic-exit and stop-failure repetition into helpers - Use Promise.allSettled in SSH transport teardown - Use the shared errorMessage() instead of inline copies - Fold isLiveSshPtyLease into ssh-types and drop the duplicate predicate Co-authored-by: m4air <m4air@Mac.localdomain> |
||
|
Jinwoo Hong
|
a8ce574c77 |
Add inert layout commands and runtime transitions (#26749)
* feat(layout): add the inert workspace layout model, Loader and Serializer One model of a workspace's tab layout (all tab kinds, groups, split trees, pane bindings, titles/colors/pin, sleeping records) in src/shared/workspace-layout, with a Loader from today's stored session partition and a Serializer back to it. The structural rules check and the pure pane/group tree helpers move into the same folder so the module can run them; their callers import the new paths. Nothing in the app calls the new module yet. * feat(layout): add inert layout commands and runtime transitions applyLayoutCommand applies every layout command a client can send (create, split, close, move, reorder, rename, recolor, pin, open editor/browser/agent tabs, start/restart, sleep/wake) to the layout model, and applyLayoutTransition applies the facts the runtime sees (process started or exited, SSH lease ended, orphan adopted, owner removed or renamed, agent launch verdict, sleeping records). Pure tab-order, group-move and pane-split helpers move into the shared module so the renderer, today's headless host and the new commands share one implementation. Nothing in the app calls the new commands. * fix(layout): avoid toSorted in shared code and drop an unused helper The mobile bundle includes src/shared and Hermes lacks Array#toSorted. * refactor(layout): name tree comparisons for what they ignore * fix(layout): one stored home per layout fact; Loader reports every rule it applies The row's terminal id, preview flag, worktree id and execution host are now derived on save instead of kept a second time. Groups never stay empty. The Loader reports row/tab-bar disagreements, re-minted ids, worktree id and host disagreements, keeps a row stored in two workspaces where the tab bar names it, and gives a pane id repeated in a second tab a new unbound pane. The rules check runs on what the Serializer writes. Fixtures become test fixtures, and a test lists every model field with the one fact it holds. * fix(layout): commands keep one home per fact and refuse instead of throwing createGroup is gone (a group only exists while it holds a tab); a lone tab split beside another group moves its group leaf and keeps the id; moveTab refuses when the move cannot be built; renaming onto an existing workspace is refused; worktree and execution host ids come from the workspace. * fix(layout): derive each tab's host from its owner and keep one preview fact per editor tab An editor tab names its file's owner (an SSH target or paired runtime), every other tab its partition. Where the stored tab and file preview flags disagree the tab loads permanent, so a draft can never be replaced. A row terminal that differs from the one derived from its focused pane is now reported. * fix(layout): one binding kind, one title, one host rule; records follow a reassigned pane A legacy row with no pane layout now loads as a one-pane tab bound to its terminal (reusing the leaf its records name), so the null-panes branches and legacyPtyId are gone. A pane given a new id because two tabs shared it takes its sleeping and incarnation records, scrollback and focus with it. A terminal's live title is one fact (the tab-bar label wins on load). Every tab-bar entry names its execution host by one rule. The fact-location test is now keyed by on-disk field and checked by mutating every source. * fix(layout): commands drop the legacy binding branches; rename carries closed-tab records Every terminal tab has panes now, so commands lose their null-panes branches and a close always releases the tab's terminals. The pane-record rekey is shared with the Loader. Renaming a workspace renames its closed-tab records' worktree id, so a later removal or partition move still selects them. * fix(layout): one change report by construction; records without copies The Loader now picks values by fixed precedence only and reports one field-by-field diff of the stored session against what the next save writes, over the disk-field tables (now in workspace-layout-disk-fields.ts, with session fields derived from SESSION_FIELD_OWNERS). Sleeping and closed-tab records live in their workspace without the pane key, tab and worktree id their key and workspace already give; the Serializer fills them. Tab-bar entries are written in the one tab order. * fix(layout): drop transitions with no producer yet; group moves hold no selection orphanAdopted, legacyWorkerRecovered, clientHostedPageAnnounced, agentLaunchVerdict, sleepingRecordsCaptured and moveWorkspaceToPartition ship with the PRs that produce them. The shared group moves take and return LayoutGroup only; the headless host keeps its selected tab beside them. Commands keep sleeping and closed-tab records in their workspace, so rename and removal need no record patching. * fix(layout): name the change report's key-list parameter's type * refactor(layout): key pane data by leaf, check the model itself, one legacy record - Each pane's data (terminal, incarnation, title, sleeping agent) is one leaf record keyed by leaf id, and scrollback beside it is keyed the same way. Pane keys are derived only when saving. The Loader gives a repeated leaf id a new id before attaching records, so rekeyPaneRecords is gone; records naming no pane are carried as stored. - The rules check runs on the model directly (ungrouped tab, empty group, group tree that disagrees with the groups, pane data without a pane); the disk-format rules stay separate. - The row-ownership marker and topology revision move into one legacy record outside WorkspaceLayout, updated by updateLegacyPersistence after each apply. - Fixtures include a terminal tab whose tab-bar id and terminal id differ. * refactor(layout): commands on leaf-keyed pane data and the one legacy update - Commands and transitions read and write each pane's one leaf record; drag-out moves only the tree, so the pane's data and scrollback follow by structure. No command advances the topology revision itself: applyLayoutCommand and applyLayoutTransition run updateLegacyPersistence after each apply, which also marks a workspace as owning its rows once it holds a terminal. - Preview replacement closes the preview through closeTabs, so an unsaved draft refuses it as any close does. - Tests: create then close then save with and without a reload, commands on a tab whose tab-bar and terminal ids differ, drag-out scrollback, the preview refusal. * fix(layout): join the group ids the tree-mismatch detail names * fix(layout): leave the stored revision map as is when no repo advances |
||
|
OrcaWinandm4air
|
cce31b4b4d |
refactor(renderer-terminal-runtime): clean up Phase 3-3.8 code (#26900)
- One helper plans and runs a paste into a terminal pane (five copies before) - Reuse getShortcutPlatform instead of three private copies - runtime-git-client becomes plain re-exports - TerminalErrorToast builds each token's test/replace regex from one helper - Reuse PreconnectInputWriter for the recovery input hold - Name the session-tabs snapshot callback type once - Narrow isTerminalDriverState with 'in' instead of a cast - Inline the single-use terminal-pane-client-host wrapper Co-authored-by: m4air <m4air@Mac.localdomain> |
||
  Willian FelipeandNeil
|
150bf49cb2 |
feat(kiro): report Kiro agent status and list Kiro sessions in history (#26695)
* feat(kiro): report Kiro agent status through managed hooks Kiro keeps hooks inside each agent config, so Orca installs one managed hook per event into $KIRO_HOME/agents/*.json (locally and on SSH hosts under the remote KIRO_HOME) and normalizes the payloads on /hook/kiro. Pets, sidebar, notifications and mobile now see Kiro working and done. Kiro resume is gated on a new agent-session.kiro-resume.v1 capability. * feat(kiro): list Kiro sessions in Agent session history Reads $KIRO_HOME/sessions/cli/<id>.json metadata plus the streamed <id>.jsonl transcript locally, under WSL and over SSH, skips subagent child sessions, and resumes with kiro-cli chat --tui --resume-id <id>. * fix(kiro): report a remote agent deleted mid-install as unreadable * fix(kiro): resume history rows by the validated session file name * fix(kiro): preserve hook configs and restore sessions correctly * test(kiro): use typed filesystem callbacks * test(kiro): preserve older-peer search expectations --------- Co-authored-by: Neil <neil@stably.ai> |
||
|
Jinwoo Hong
|
9021133929 |
Add the inert workspace layout model, Loader and Serializer (#26748)
* feat(layout): add the inert workspace layout model, Loader and Serializer One model of a workspace's tab layout (all tab kinds, groups, split trees, pane bindings, titles/colors/pin, sleeping records) in src/shared/workspace-layout, with a Loader from today's stored session partition and a Serializer back to it. The structural rules check and the pure pane/group tree helpers move into the same folder so the module can run them; their callers import the new paths. Nothing in the app calls the new module yet. * fix(layout): avoid toSorted in shared code and drop an unused helper The mobile bundle includes src/shared and Hermes lacks Array#toSorted. * fix(layout): one stored home per layout fact; Loader reports every rule it applies The row's terminal id, preview flag, worktree id and execution host are now derived on save instead of kept a second time. Groups never stay empty. The Loader reports row/tab-bar disagreements, re-minted ids, worktree id and host disagreements, keeps a row stored in two workspaces where the tab bar names it, and gives a pane id repeated in a second tab a new unbound pane. The rules check runs on what the Serializer writes. Fixtures become test fixtures, and a test lists every model field with the one fact it holds. * fix(layout): derive each tab's host from its owner and keep one preview fact per editor tab An editor tab names its file's owner (an SSH target or paired runtime), every other tab its partition. Where the stored tab and file preview flags disagree the tab loads permanent, so a draft can never be replaced. A row terminal that differs from the one derived from its focused pane is now reported. * fix(layout): one binding kind, one title, one host rule; records follow a reassigned pane A legacy row with no pane layout now loads as a one-pane tab bound to its terminal (reusing the leaf its records name), so the null-panes branches and legacyPtyId are gone. A pane given a new id because two tabs shared it takes its sleeping and incarnation records, scrollback and focus with it. A terminal's live title is one fact (the tab-bar label wins on load). Every tab-bar entry names its execution host by one rule. The fact-location test is now keyed by on-disk field and checked by mutating every source. * fix(layout): one change report by construction; records without copies The Loader now picks values by fixed precedence only and reports one field-by-field diff of the stored session against what the next save writes, over the disk-field tables (now in workspace-layout-disk-fields.ts, with session fields derived from SESSION_FIELD_OWNERS). Sleeping and closed-tab records live in their workspace without the pane key, tab and worktree id their key and workspace already give; the Serializer fills them. Tab-bar entries are written in the one tab order. * fix(layout): name the change report's key-list parameter's type * refactor(layout): key pane data by leaf, check the model itself, one legacy record - Each pane's data (terminal, incarnation, title, sleeping agent) is one leaf record keyed by leaf id, and scrollback beside it is keyed the same way. Pane keys are derived only when saving. The Loader gives a repeated leaf id a new id before attaching records, so rekeyPaneRecords is gone; records naming no pane are carried as stored. - The rules check runs on the model directly (ungrouped tab, empty group, group tree that disagrees with the groups, pane data without a pane); the disk-format rules stay separate. - The row-ownership marker and topology revision move into one legacy record outside WorkspaceLayout, updated by updateLegacyPersistence after each apply. - Fixtures include a terminal tab whose tab-bar id and terminal id differ. * fix(layout): join the group ids the tree-mismatch detail names * fix(layout): leave the stored revision map as is when no repo advances |
||
|
Jinwoo Hong
|
e5412cccec |
test(e2e): headless layout oracle covers tab moves between groups (#26857)
Two headless scenarios drive session.tabs.move (split into a new group, move into an existing group, reorder, and a move that empties a group) on orcad and Electron serve, and check group order, the selected tab per group, and a warm restart. The temp repo is resolved to its real path so macOS runs list its worktree without a TMPDIR workaround. Two new main findings are listed as known: headless saves groups without a tab bar, and move-to-group leaves the target group's previous selection. |
||
|
Jinwoo Hong
|
6013d093e2 |
fix(mobile): one host sidebar on iPad, with the page owning the host area (#26761)
* feat(mobile): let the page own the wide host area, one sidebar only
On wide layouts a detail route's page session ran app/h/_layout.tsx inside
the detail pane, judged itself wide, and drew a second host sidebar beside
the native one.
- The page draws the host sidebar only when init carries the new optional
`ownsHostArea` fact, never from its own width.
- The page manifest's `/h/[hostId]` route declares `canOwnHostArea`.
- On wide layouts the native host route mounts one host-area page session.
A declaring bundle serves it with every served route's grants, so hops
stay in the page, and the native layout drops its sidebar. Without the
declaration, or without a bundle, the session answers native-route and
every wide detail route stays native.
- The owner is recorded per host so the layout and switches agree and a
relaunch does not flash.
Claude-Session: lane-ipad-page-owns-host
* fix(mobile): decide the wide host area per session, without a stored owner
Round 1 kept a per-host owner record that the native layout trusted. When
the record said the page owned the area, a detail route opened natively
(deep link, notification, handoff, narrow-to-wide with a detail pushed)
showed no sidebar at all.
- Drop the record. A wide session decides from the manifest it already
read: it is served only when the host route declares canOwnHostArea,
and it rebuilds when the layout class changes.
- The native layout hides its sidebar only on the host route, while that
host's page session is serving, as the shell screen reports through a
layout context. Detail routes keep the native sidebar, and a page there
draws none, so there is exactly one.
- Phones read and render nothing new.
- The host-area page can read and write every workspace-scoped key of its
host (chat tabs, live-input handles), as a per-workspace session does.
Claude-Session: lane-ipad-page-owns-host
* test(mobile): type the layout-class doubles rather than assert them
Claude-Session: lane-ipad-page-owns-host
* fix(mobile): keep detail pages through a layout flip, guard capped keys
- A wide/narrow flip rebuilds a session only on the host route or under a
bundle without canOwnHostArea, where the answer changes. A detail page
under a declaring bundle keeps its document, scroll and draft.
- In the host-area page, a workspace key the 32-entry init cap leaves out
is named in storageOversize, so the page cannot replace the device's
chat-tab list from a default. The init frame bounds that list to the
page's schema; the host still refuses every key in the full list.
- A leaving host-area session clears only its own report.
Claude-Session: lane-ipad-page-owns-host
* refactor(mobile): host-area ownership from the reducer, storage read on demand
- The serving report is a focus-scoped boolean, so a host pushed over
another no longer leaves the first one's area unclaimed after the pop.
- The reducer takes `layout-changed {wide}`: it keeps the session when the
route view is unchanged and restarts it otherwise; `routeViewOf` returns
`ownsHostArea`.
- One web override, `usePageOwnsHostArea`; the native rule is inline in the
host layout.
- A host-area page reads a host workspace key from the shell on demand
(`native.storage.read`, granted to that session only) instead of init
enumerating the store.
Claude-Session: lane-ipad-page-owns-host
* fix(mobile): host-area page reads back its own workspace-key writes
A page read right after a page write of the same key is answered from the
page's own value, not from a shell that may not have the notify yet. A
refused read crosses as an error, never as an unset key.
Claude-Session: lane-ipad-page-owns-host
* refactor(mobile): read-your-writes from the shell's mirror, final review fixes
- The shell answers a workspace-key read from the mirror, which every
shell-side write updates synchronously (removals held as known-absent),
and reads the store only on a miss. The page shim's written-keys set
goes.
- `onLayoutChanged` compares the route view field by field; the session
is created with its layout and the layout effect fires on `wide` alone.
- The device verb dispatcher's type excludes the storage read.
- The sidebar matrix mounts the real shell screen to decide the report.
Claude-Session: lane-ipad-page-owns-host
* feat(mobile): host-area page takes the whole screen from the download on
The native sidebar steps aside once the manifest says the page owns the
host area: fetching, activating, ready and the failure screens after them
are full screen, as on a phone. It stays while the host is being checked.
Claude-Session: lane-ipad-page-owns-host
* fix(mobile): host-area workspace-key read survives a write during it
On a mirror miss the read went through the mirror's hydrate, which drops
its answer when any mirrored write lands meanwhile, so a stored key read
as unset right after a relaunch. A miss now reads the store directly, and
a store failure rejects rather than reading as unset.
Claude-Session: lane-ipad-page-owns-host
* fix(mobile): a wide detail session hands the host route to the shell
A session opened on its own under a page that can own the host area
kept a hop to the host route in-page (a missing worktree's notice, or
leaving the session), drawing a second host list beside the native
sidebar. Its pairs now leave the host route out, so the hop goes to the
shell, which opens the area-owning session. Phones keep the hop in-page.
Claude-Session: lane-ipad-page-owns-host
* test(mobile): a phone mount's layout dispatch leaves the session alone
Claude-Session: lane-ipad-page-owns-host
* fix(mobile): leaving a wide detail session returns to the host area
The leave-session chevron and the missing-worktree bounce hand the host
route to the shell on a wide layout, which pushed it: a second host-area
session over the session, with Back reopening the session. On a wide
layout the shell now dismisses to this host's route instead, taking the
href's params, or replaces the session when nothing is below it. Phones
still push.
Claude-Session: lane-ipad-page-owns-host
* fix(mobile): a narrow-wide flip restarts a detail session's page
The route view comparison now counts the paired routes, so a detail
session under a page that can own the host area restarts on a flip
instead of keeping the narrow pairs, whose host-route pair let the page
draw a second host list beside the native sidebar.
Claude-Session: lane-ipad-page-owns-host
|
||
|
|
1f8dce85d4 |
refactor(runtime-rpc-terminal-io): clean up Phase 3-3.8 code (#26895)
* refactor(runtime-rpc-terminal-io): clean up Phase 3-3.8 code - One admission prologue and one unary invocation shared by both RPC dispatchers - Delete rpc schema re-export shims; import from the shared params catalog - Delete the unregistered duplicate markdown tab method module - Collapse multiplex installer stage types; add isMultiplexStreamAttached - One source-identity guard in the ACK recovery snapshot - Single stream teardown in the renderer multiplexer - De-duplicate RPC method declarations; derive dispatch options from RpcContext - Share viewport-frame decoding and one head-trim loop for pending queues - Single sendTerminal call in sendTerminalStreamInput * test: point cross-version and scope checks at the moved browser tab-create schema; run transport parity test under Node --------- Co-authored-by: m4air <m4air@Mac.localdomain> Co-authored-by: m4air <m4air@m4airs-MacBook-Air.local> |
||
|
OrcaWinandm4air
|
fbcf0edc60 |
refactor(pty): one remote-runtime PTY id predicate (#26930)
Co-authored-by: m4air <m4air@m4airs-MacBook-Air.local> |
||
|
OrcaWinandm4air
|
54b623b6a3 |
refactor(execution-host): one connection-to-host helper (#26929)
Co-authored-by: m4air <m4air@m4airs-MacBook-Air.local> |
||
|
cbc6f54e1c |
refactor(runtime-core): clean up Phase 3-3.8 code (#26894)
* refactor(runtime-core): clean up Phase 3-3.8 code - worktree.ps: summary lookup is a closure, not a 4-arg callback - relay: shared RelayStreamAckWindow for fs and git response streams - relay: terminal-frame slots live on RelayStreamRegistry - one isBinaryBuffer/BINARY_PROBE_BYTES in shared/binary-buffer - one automation owner fence on the controller, no casts - drop unchecked Store cast in remote repo registration - speech download reuses the shared abort-aware delay, no casts - RelayRuntimeServices: drop leftover handler list and param casts - use shared errorMessage() in audited files Co-Authored-By: Claude <noreply@anthropic.com> * test(cross-version-wire): pass worktree ps summary lookup in both builds' shapes --------- Co-authored-by: m4air <m4air@Mac.localdomain> Co-authored-by: Claude <noreply@anthropic.com> Co-authored-by: m4air <m4air@m4airs-MacBook-Air.local> |
||
|
Jinwoo Hong
|
3eea6a5368 |
fix(mobile): size onboarding slides to the modal sheet on iPad (#26919)
* fix(mobile): size onboarding slides from the modal viewport * fix(mobile): ignore zero-width onboarding viewport layouts * test(mobile): model slide progress in onboarding width test |
||
 github-actions[bot]
|
5da51ea6b2 | Update README downloads badge | ||
 NeilandJoaquin Terrasa
|
d2d17a3e47 |
fix(orcad): show help before server startup
Make orcad --help print usage and exit before native preflight or server startup, preserving existing flag parsing and managed-runtime startup. Verified against the built server and launcher. Adapted from #26621 by @espetro. Co-authored-by: Joaquin Terrasa <josocjoq+dev@pm.me> |
||
 NeilandJustas Brazauskas
|
65580c5315 |
Honor OpenCode 2 launch mode when resuming sessions (#26843)
Let the configured launch command own the server mode when resuming OpenCode 2 sessions. Resume arguments select only the session, while Session History uses the existing default launch command so default standalone behavior is preserved. Fixes #25003. Retains the contribution from #25004. Co-authored-by: Justas Brazauskas <brazauskasjustas@gmail.com> |
||
|
Neil
|
077b62dee0 | fix: use alternate browser for rich Markdown modifier clicks (#26920) | ||
|
Neilandm4air
|
7d24509000 |
Reject terminal file drops from a different execution host (#26830)
A workspace file dragged into a terminal now pastes only when its source host (and paired-server namespace) matches the receiving terminal's transport; otherwise the drop is refused with a toast. Co-authored-by: m4air <m4air@m4airs-MacBook-Air.local> |
||
|
OrcaWinandm4air
|
1b2d07591d |
refactor(ssh-connect-managed): clean up Phase 3-3.8 code (#26890)
Remove the dead connectViaSystemSsh startup path, collapse SshConnection's repeated transport teardown and duplicated passphrase rung, share one execHostCommand across relay install modules, share the orcad preflight short-circuit, single-source telemetry pick/libc, drop dead exports and the TCP_NODELAY socket probe, and trim sftp-upload and reconnect-classifier duplication. No behavior change. Co-authored-by: m4air <m4air@Mac.localdomain> |
||
|
Neilandm4air
|
51e7181850 |
Keep remote terminal path links on their own host (#26696)
* Keep terminal workspace root links on their execution host * Resolve terminal root-link source host from the owner, not a missing connection id A direct SSH pane whose worktree id also exists on the desktop has no connectionId (ambiguous owner), which was read as local and offered the desktop workspace. Resolve the source via the owner-route resolver and skip the root shortcut when the source host is unknown. * Resolve terminal file links against the source owner's host A same-id SSH workspace left connectionId unset, so a rejected root link, the default-app row and Shift-click fell through to this machine's files. The terminal file context now names the resolved direct SSH host and refuses file access when no owner can be determined. * Keep an ambiguous terminal link owner unresolved instead of local The connection lookup ignores detected worktree rows, so a registered local row could override an owner the full resolver found ambiguous (registered local, detected SSH) and send link stat/open to this machine. Only a missing route may now defer to the connection lookup. * Re-resolve a held chat link's owner before rechecking it Native chat rechecks held link targets when a turn ends. A target resolved while its owner was local kept that context after a same-id worktree was detected on an SSH host, so the recheck still asked this machine. Each existence ask now re-resolves the workspace owner and refuses when it is unresolved or names a different host. --------- Co-authored-by: m4air <m4air@Mac.localdomain> |
||
|
|
5f62f498f3 |
refactor(orcad-daemon-startup-cli): clean up Phase 3-3.8 code (#26892)
* refactor(orcad-daemon-startup-cli): clean up Phase 3-3.8 code - One helper (getAllDaemonAdapters / listEveryDaemonGeneration) for enumerating daemon generations, replacing five hand-rolled copies. - Shared serve pairing/endpoint readiness for Electron serve and orcad. - One option list builds both --serve-* and orcad serve argv. - holdDaemonAdoptionLease takes an options object; reuse childProcessHasExited. - Drop orcad-entry's dead re-exports and repeated version fallback. - Reuse the shared withTimeout in daemon retirement. - Share the install-wait step in the serve update supervisor. - Trim stale or misplaced comments. Behavior unchanged; production code is net -132 lines. * test(serve): keep real serveOptionArgs in launch mock; pin Electron serve argv --------- Co-authored-by: m4air <m4air@Mac.localdomain> Co-authored-by: m4air <m4air@m4airs-MacBook-Air.local> |
||
|
Brennan Benson
|
7b91405547 |
Let in-page links switch sections inside a chat visual (#26729)
* Let in-page links switch sections inside a chat visual * Make in-page links work in chat visuals on mobile and in WebKit The first fix (an about:srcdoc base in the visual) only reached desktop Chromium. On mobile, the page hosting the visual carried a policy with base-uri 'none', and a srcdoc visual inherits its host's policy before its own, so the base was refused and an in-page link still took the visual off its page and showed it as failed. The mobile web shell sends the same rule as a header. Separately, WebKit (iOS, and the web client in Safari) fires the frame's load event for an in-page link inside a sandboxed frame, and both hosts treated any second load as the visual navigating away and retired it. - The mobile host page and the mobile web shell header now allow only an about: base, so the shell's own base applies; the visual's own policy still refuses any base it declares. - Instead of retiring a visual on its second load, the host now asks the visual's shell after every load. A later load must be answered by the same run of the shell within the plugin panel watchdog's 5 s deadline; no answer, or an answer from a new run (a reload), still retires it. - Tests: unit tests for the policy, the answer reader and the frame's keep/retire rules, and a Chromium + WebKit test of both mobile surfaces that runs in the mobile web app CI job. * Keep any answered later load, and say plainly what the check is for The previous commit tagged each run of a visual's wrapper with a random id so a reload would still retire it. The visual shares a window with its wrapper, so it could read or set that id itself, and a reload only reruns the same Orca-written document, which is not an escape. Drop it: after any load past the first, the host asks once, and any answer on this visual's channel within the wait keeps it; silence retires it. - Comments now say the check is a tripwire for a frame gone blank or to another page; the frame guard (desktop), the host page's policy plus the app's load filter (mobile) and the sandbox are the boundaries. - Visuals get their own 5 s wait instead of borrowing the plugin panel watchdog's, because a 1 s wait retired a busy visual in WebKit. - Tests: a reload is now kept; going blank, or dropping the base and following a link, still retires. The CI routing for the shared wrapper is now pinned in the path-classification test. * Load the visual links check's mobile modules only where they can build The unit shards run every config/scripts test with only the root dependencies installed. This check imported a mobile/ module at load, and transforming it reads mobile/tsconfig.json, which extends a package those shards do not have. The transform error could not be sent back to the test worker under Bun, so the whole shard stopped with "DataCloneError: The object can not be cloned". Import the visual shell and the mobile host page inside the dependency gate, as the session-dictation check already does, so the shards skip the file without loading them. The mobile web app job, which installs mobile dependencies, still runs it in full. |
||
|
OrcaWinandm4air
|
7e621d5a4a |
refactor(ssh-orcad-lifecycle): clean up Phase 3-3.8 code (#26891)
Co-authored-by: m4air <m4air@Mac.localdomain> |
||
|
Jinwoo Hong
|
94ed1675e3 |
refactor(mobile): let the build alone decide the OTA shell (#26825)
Remove the Troubleshoot hybrid shell toggle and its stored flag. The shell mounts iff the binary was built with EXPO_PUBLIC_MOBILE_SHELL=ota, decided synchronously, so route switches no longer need a pending frame. |
||
|
|
1c773a3e97 |
refactor(ssh-relay-runtime): clean up Phase 3-3.8 code (#26889)
- Pinned-relay planning throws its PinnedRelayFallbackError instead of returning a host-node plan - One shared execHostCommand replaces eight private copies - Install lock shares its post-acquire GC recheck and unconfirmed-termination catch - Relay session tracks unsupported optional methods in one set - Previous-relay census carries one status instead of two booleans - Remote-node resolution drops the signal-only option shims Co-authored-by: m4air <m4air@Mac.localdomain> Co-authored-by: Claude <noreply@anthropic.com> |
||
|
Neil
|
bd85767ee7 | [CI] Use public threads for mobile unit tests (#26899) | ||
|
OrcaWinandm4air
|
40716d702e |
fix(remote-terminal): keep the disconnected-input grace honest across retries (#26871)
* fix(remote-terminal): keep the disconnected-input grace honest across retries A web pane's Reconnect cancelled recovery to idle before replaying its attach, which reset the 5-minute grace, so each failed Reconnect extended how long held keys could wait. And a retry started inside the grace discarded every held key once the grace ended, including keys typed during that retry. Held input is now split into cohorts at each retry from the latch: the sealed cohort keeps the grace of the latch it was held through, keys typed during the retry are only charged to a later latch, and the web Reconnect cancel no longer resets the grace. * fix(remote-terminal): a failed in-flight write from an expired cohort keeps newer input Dropping an expired cohort discards the journal, which fails a held accepted write still in flight; the suspended flush then treated that as a broken link and cleared the keys typed during the retry. The dropped write is now retired, so the flush pauses or continues with the newer cohort instead. * fix(remote-terminal): keep held input whole across a retry begun inside the grace Dropping an expired older part of the hold while delivering keys typed after it is unsafe: held input is one ordered stream, and the dropped part may be the Ctrl+C the newer keys rely on. Replace the cohort split with an order-safe bound: a retry begun inside the grace carries the whole hold until it ends, and the next latch measures from the original start, so the overrun is at most one recovery window. * fix(remote-terminal): replay a web pane Reconnect inside a recovery epoch The web Reconnect path cancelled recovery to idle before replaying its attach, which cleared the deadline: an attach whose snapshot stalled could release held input at any later time. It now begins a fresh recovery epoch, as auto-recovery replays do, so the deadline bounds the attempt and fences a late snapshot, and the cancel no longer counts as a recovery that resets the grace. * docs(remote-terminal): grace reset wording matches its reset points --------- Co-authored-by: m4air <m4air@Mac.localdomain> |
||
|
Neil
|
9d1cd1e8b1 |
[CI] Consolidate related tests and advance transcript waits with the existing virtual clock (#26874)
* [CI] Consolidate related tests and advance transcript waits * [CI] Refresh reliability gate references after test consolidation |
||
|
OrcaWinandm4air
|
3f37f7ea6d |
Keep an unplaceable row owner out of desktop Open in launches (#26797)
Co-authored-by: m4air <m4air@Mac.localdomain> |
||
 Jinjing
|
c9c6a00449 |
Keep usage overflow chip width stable as hidden count changes (#26873)
The "+N" overflow chip in the status bar changed width whenever its count changed, so the density probe that decides how many usage chips fit kept re-measuring. Reserve the width of the provider-count label with an invisible sizer span, so the chip keeps the same size across counts. - Pass the total provider count into UsageOverflowChip and render it in an invisible, grid-stacked sizer span. - Overlay the visible "+N" label on the same grid cell so the chip width tracks the total, not the hidden subset. - Add an e2e test that loads ten usage providers, shrinks the viewport to collapse them, and checks the chip width stays stable and no "Maximum update depth" or React #185 errors occur. |
||
|
Neilandm4air
|
05ab94bd65 |
Move shared file watches when the selected remote host changes (#26632)
* Invalidate shared file watch targets when the selected host changes * Watch the selected host's own workspace root when hosts share a workspace id * Watch each open editor on its own execution host instead of the selected host --------- Co-authored-by: m4air <m4air@Mac.localdomain> |